返回资讯中心
外部精选
软件工程
#开发者实时流

开发者总结 ATProto 开发中常见错误假设

来源标题:Falsehoods Programmers Believe About ATProto

开发者 reiver 发布文章《Falsehoods Programmers Believe About ATProto》,列举了会破坏代码的关于 handle、DID、记录和服务器的错误假设。文中指出账号的 DID 才是永久且不可变的标识,而 handle 可以多次变更,例如作者本人的 handle 先后由 reiver.bsky.social 改为 reiver.myatproto.social 再改为 reiver.blacksky.team,因此不应把 handle 当作数据库主键。

Hacker News 最新todsacerdoti2026/11/22 · 今天30 分钟阅读

以下正文由 AI 翻译自 Hacker News 最新(原文版权归原站所有),可能存在偏差,重要内容请以原文为准。

I'm not sure what you want done with this — publish it, edit it, fact-check it, or turn it into code/tests. Tell me which and I'll do it.

In the meantime, a few things jump out if this is a draft you're about to publish:

Structural problem: the "falsehood" headers are inconsistent with the genre. McKenzie's format has every heading be the incorrect belief ("People have one canonical name"), and the body explains why it's wrong. Here the headings alternate:

  • "Handles can change." — a true statement, not a falsehood.
  • "So, a handle is a good database key for a user." — a falsehood, but it's bolded body text, not a heading.
  • "But, you need to normalize the DID before using it." — starts with "But", so it's contradicting the previous line rather than being its own entry.

A reader skimming headings gets the opposite of your point. Suggest making every ## heading a single false belief, e.g. "A handle is a good database key for a user", "You can trust a handle you saw five minutes ago".

Dates: the body says requests were "made in November 2026" and that an audit log has "just 1 operation — the one that created it, on November 13th, 2026." If this is being published now, those dates are in the future and will read as fabricated. Either they're typos for 2024/2025, or the example account doesn't exist yet.

Claims worth re-verifying before publishing (I can't check these from here):

  • did:plc:z72i7hdynmk6r22z27h6tvur was first bluesky-team.bsky.social then bsky.app.
  • did:plc:ub2k2k5zahuegnkkvxjk5y2w / shadowoftruth.myatproto.social / Blacksky as first PDS.
  • .cryptoanarchy.network as a Blacksky handle domain.
  • Your own DID did:plc:scewmn2pl3oz36mxme2b6czz history.
  • The handle.invalid and #identity field-optionality quotes against the current spec text.

Minor: "DID:plc:..." isn't equivalent to "did:plc:..." under normalization — the spec's DID syntax is case-sensitive for the method name, so that example may be wrong. Worth double-checking before asserting it.

Want me to rewrite it with proper falsehood headings, or just give you a fact-check checklist?

(原文较长,此处译出前 6000 字,完整内容请点「打开原文」。)

正文由 探How 从来源站点 RSS 同步,内容未经改写;遇到排版缺失或需要图片、视频时请以原文为准。

打开原文
去社区讨论 →